Accurate feed evaluation is important for feed formulation in aquaculture. In this study a net energy (NE) evaluation method was tested which recognises differences in the utilisation of the different macro-nutrients (protein, fat and carbohydrates) for growth. Therefore, a feeding trial was performed with feeds that differed in energy level and with a wide contrast in macronutrient composition, reflecting those found in practical rainbow trout diets. The trial was set up in a 2 by 4 factorial design which included two protein levels (CP400 and CP500) and four different fat levels (CF140, CF190, CF240, CF290) coinciding in a wide range in carbohydrate levels. These differences were achieved by starting from a basal high protein and fat diet where protein and or fat were replaced by carbohydrates. These 8 diets were then used to assess the prediction precision of growth performance by a NE approach as compared to a digestible energy (DE) approach in feed evaluation for rainbow trout. The results of the current study showed that replacing fat by carbohydrates reduced the utilisation of energy for growth. This effect was stronger at the highest carbohydrates levels which indicated a curvilinear response in the utilisation of carbohydrates. However in terms of the prediction precision of growth rate an improvement was only shown when using a NE approach using linear efficiencies of digestible macronutrients (R2 = 0.890 compared to R2 = 0.848 for the DE approach). A NE approach assuming a curvilinear effect of carbohydrates did not lead to an improved prediction of growth rate in the current study (R2 = 0.617).